Crypto, AI and the Permanent Underclass | The Breakdown

What It Covers

The episode examines how crypto offers direct investment exposure to emerging financial primitives like AMMs and liquidity mining, unlike the internet era where investors could only access infrastructure companies like Cisco. It explores whether AI agent technology will produce comparable on-chain primitives that could provide economic opportunity amid AI-driven job displacement concerns.

Key Questions Answered

• •Primitive Investment Velocity: Crypto enables immediate exposure to foundational building blocks through token purchases, contrasting with internet-era investing where access required waiting for infrastructure company IPOs. Early ETH investors who recognized the ERC-20 standard's potential before the 2017 ICO boom saw 2,200x returns from 68 cents to $1,500, demonstrating how identifying primitives before mass adoption creates asymmetric opportunities unavailable in traditional technology investing cycles.
• •Liquidity Mining as Reusable Pattern: Compound's 2020 model rewarding both borrowers and lenders with COMP tokens created a composable primitive that Balancer, Uniswap, and Yearn replicated. This drove Ethereum TVL from under $2 billion to nearly $10 billion in three months and contributed to ETH's 20x price increase through 2021. Recognizing when protocols popularize new incentive structures signals upcoming ecosystem-wide adoption waves worth front-running.
• •AI Infrastructure Lacks True Primitives: Current crypto-AI projects like Akash, Render, and Helium tokenize off-chain services rather than creating novel financial primitives. Only Filecoin introduces genuinely new cryptographic primitives through storage proofs that verify commitment over time. GPU marketplaces and bandwidth networks apply existing primitives like tokens and staking to new markets without establishing reusable on-chain building blocks comparable to AMMs or ERC-20 standards.
• •Agent Identity Standard Potential: ERC-8004 token standard could give AI agents portable on-chain identities with attached attestations and reputation scores, addressing trust problems when billions of agents conduct autonomous economic activity. Combined with Coinbase's x402 protocol for stablecoin payments over HTTP and DeFi yield opportunities, these components might form an investable primitive stack if standardized, though OpenClaw founder Peter Steinberger currently rejects crypto integration despite open-source ecosystem emergence.
• •Privacy Protocols Enable Data Access: Foundation models will exhaust public web data by decade's end, requiring access to company-held deep web data estimated at 100x public web volume. Fully homomorphic encryption protocols like Zama, along with zero-knowledge and multiparty computation solutions, provide privacy rails necessary for AI agents and robotics to access proprietary training data. Investing in these privacy infrastructure protocols positions for long-term AI development needs beyond current hype cycles.